Key results
The company
Falabella
falabella.comMultinational department store and e-commerce retailer in Latin America.
Result highlights
- Ticket resolution time cut from 2 hours to 30 mins
- Ticket creation time cut from 16 mins to 4 mins
- Ticket categorization accuracy improved to 80%
- 33% reduction in manual ticket creation
The story
Latin America's leading digital banking and retail ecosystem serves 36 million customers across seven countries with a workforce of 90,000 employees.
Opening a support ticket could take up to 10 minutes while retail customers waited at checkout, creating critical operational friction. Informal reporting channels and manual processes resulted in only 40% categorization accuracy for user-created tickets.
The organization deployed a multi-channel AI agent accessible via Voice, Web, Teams, and WhatsApp to standardize incident reporting across the region. A multi-agent architecture powered by Gemini models captures incident details and automatically creates categorized tickets without human intervention. Vertex AI Search implements a retrieval-augmented generation (RAG) pipeline to ground these interactions in internal process documentation.
Scope & timeline
- 22,000+ tickets processed via AI agent
Quotes
“Abrir un ticket o llamar a la mesa de ayuda podía demorar hasta 10 minutos, mientras se tenía a una persona esperando en una caja. Eso es fatal para el sector de retail. Si la persona tiene que esperar para comprar, se irá a otro lado”